Dual modifying of MAVS at lysine 7 by SIRT3-catalyzed deacetylation and SIRT5-catalyzed desuccinylation orchestrates antiviral innate immunity

Significance

As the first line of defense against viral infection, innate immunity must be precisely controlled. Mitochondrial antiviral signaling protein (MAVS) is a critical adaptor protein in the retinoic-acid inducible gene I (RIG-I)-like receptor (RLR) signaling pathway. In this study, we found that sirtuin3 (SIRT3) interacts with MAVS to catalyze MAVS deacetylation at lysine residue 7 (K7), thereby promoting MAVS activation, in contrast to SIRT5-mediated MAVS desuccinylation at K7. Loss of Sirt3 renders the host more susceptible to viral infection compared to its wild-type siblings. However, Sirt3 and Sirt5 double-deficient mice exhibit the same viral susceptibility as their wild-type littermates. Our results suggest that SIRT3 serves as an accelerator and SIRT5 as a brake to orchestrate antiviral innate immunity.

Abstract

To effectively protect the host from viral infection while avoiding excessive immunopathology, the innate immune response must be tightly controlled. However, the precise regulation of antiviral innate immunity and the underlying mechanisms remain unclear. Here, we find that sirtuin3 (SIRT3) interacts with mitochondrial antiviral signaling protein (MAVS) to catalyze MAVS deacetylation at lysine residue 7 (K7), which promotes MAVS aggregation, as well as TANK-binding kinase I and IRF3 phosphorylation, resulting in increased MAVS activation and enhanced type I interferon signaling. Consistent with these findings, loss of Sirt3 in mice and zebrafish renders them more susceptible to viral infection compared to their wild-type (WT) siblings. However, Sirt3 and Sirt5 double-deficient mice exhibit the same viral susceptibility as their WT littermates, suggesting that loss of Sirt5 in Sirt3-deficient mice may counteract the increased viral susceptibility displayed in Sirt3-deficient mice. Thus, we not only demonstrate that SIRT3 positively regulates antiviral immunity in vitro and in vivo, likely via MAVS, but also uncover a previously unrecognized mechanism by which SIRT3 acts as an accelerator and SIRT5 as a brake to orchestrate antiviral innate immunity.

The mitochondrial antiviral signaling protein (MAVS) (also known as IPS1, VISA, or Cardif) serves as a key hub in innate immunity against RNA viruses, localizes to the mitochondrial outer membrane (1–4). Upon RNA virus infection, MAVS forms prion-like aggregates by receiving the cytosolic retinoic acid-inducible gene I (RIG-I)-activated signal and subsequently forms a complex with TANK-binding kinase I (TBK1)/Iκ-B kinase ε (IKK-ε), leading to activation of the type I interferon signaling (2, 3, 5–7). In the resting states, MAVS is prevented from spontaneous aggregation to turn off the type I interferon signaling and maintain immune homeostasis (8).

Due to the importance of posttranslational modifications (PTMs) in regulating the activity, stability, and folding of target proteins by inducing their covalent attachment to new functional groups, PTMs on MAVS may play a pivotal role in the regulation of antiviral innate immune signaling (6, 9–22). Recently, we identified that SIRT5 desuccinylates MAVS at K7 to reduce MAVS aggregation, but its succinyltransferase is still unclear (22).

Sirtuin 3 (SIRT3) belongs to a conserved family of NAD⁺-dependent protein deacetylases involved in metabolism, stress response, and longevity (23). Among the seven mammalian sirtuins (SIRT1-7), SIRT3 is mainly localized to the mitochondria and serves as one of the most prominent deacetylases that can regulate acetylation levels in mitochondria, leading to the activation of many oxidative pathways (24–26). Some evidence suggests that the deacetylase activity of SIRT3 is associated with inflammation and suppression of virus production (27, 28), but whether SIRT3 is involved in antiviral innate immunity is poorly understood.

Here, we show that SIRT3 positively regulates antiviral innate immune responses both in vitro and in vivo, likely via MAVS.

Results

SIRT3 Positively Regulates the MAVS-Mediated Antiviral Innate Immune Response.

Upon infection with Sendai virus (SeV) in H1299 cells, overexpression of SIRT3 (Flag-SIRT3) caused a significant increase in IFNβ, CXCL10, and ISG15 mRNA compared to transfection of the empty vector control (Flag empty) (Fig. 1 A–C). In contrast, disruption of SIRT3 in H1299 cells resulted in a reduction of IFNβ, CXCL10, and ISG15 mRNA in response to SeV infection (Fig. 1 D–F). Furthermore, SeV-induced IRF3 phosphorylation was reduced in SIRT3-deficient H1299 cells (SIRT3^−/−) compared to wild-type (WT) H1299 cells (SIRT3^+/+) (Fig. 1 G and H). Consistent with these findings, SeV propagation was increased in SIRT3^−/− H1299 cells compared to that in SIRT3^+/+ H1299 cells as shown by immunoblotting for SeV protein (Fig. 1 G and H). VSV (Vesicular Stomatitis Virus)-GFP virus proliferation was increased in SIRT3^−/− H1299 cells compared to that in SIRT3^+/+ H1299 cells as revealed by fluorescence microscopy imaging and immunoblotting (Fig. 1I). Similar results were obtained in HEK293T cells or H1299 cells (SI Appendix, Fig. S1 A–F). These data suggest that SIRT3 positively regulates type I interferon signaling during RNA virus infection.

Fig. 1.

SIRT3 activates the MAVS-mediated innate antiviral response. (A–C) qPCR analysis of IFNβ (A), CXCL10 (B), and ISG15 (C) mRNA in H1299 cells transfected with indicated plasmids for 24 h, and then left uninfected (UI) or infected with SeV for 8 h. (D–F) qPCR analysis of IFNβ (D), CXCL10 (E), and ISG15 (F) mRNA in SIRT3^−/− or SIRT3^+/+ H1299 cells left UI or infected with SeV for 8 h. (G) Immunoblotting (IB) of whole cell lysates using the indicated antibodies in SIRT3^+/+ and SIRT3^−/− H1299 cells UI or infected with SeV for 4 or 8 h. (H) Quantitation of phosphorylated IRF3 (p-IRF3) and SeV proteins shown in (G). (I) SIRT3^−/− or SIRT3^+/+ H1299 cells were infected with VSV-GFP virus (MOI = 0.1) for 12 h, and viral infectivity was detected by fluorescence microscopy and immunoblotting. (J)Schematic of the working model of high molecular weight poly I:C (poly I:C-HMW) and low molecular weight poly I:C (poly I:C-LMW). (K) qPCR analysis of IFNβ mRNA in H1299 cells transfected with the indicated plasmids for 24 h, and then transfected without (UT) or with poly I:C HMW or poly I:C LMW for 8 h. (L) qPCR analysis of IFNβ mRNA in SIRT3^−/− or SIRT3^+/+ H1299 cells transfected without (UT) or with high molecular weight poly I:C (poly I:C HMW) or low molecular weight poly I:C (poly I:C LMW) for 8 h. (M) ISRE reporter activity by cotransfection of HA-MAVS together with Flag empty vector (Flag empty) or with Flag-SIRT3 in HEK293T cells for 24 h. (N) IFNβ promoter activity by cotransfection of HA-MAVS together with Flag empty vector (Flag empty) or with Flag-SIRT3 in HEK293T cells for 24 h. (O) qPCR analysis of IFNβ mRNA upon cotransfection of HA-MAVS together with Flag empty vector (Flag empty) or with Flag-SIRT3 in HEK293T cells for 24 h. (P–S) qPCR analysis of Ifnβ (P), Isg15 (Q), Cxcl10 (R), and Ifit1 (S) mRNA in Mavs^+/+ and Mavs^−/− MEF cells stably expressing empty vector or Sirt3 (lentivirus infection), followed by infection without (UI) or with VSV for 8 h. (T and U) Mavs^+/+ and Mavs^−/− MEF cells stably expressing empty vector or Sirt3 were infected with VSV-GFP virus (MOI = 0.1) for 12 h, and viral infectivity was detected by fluorescence microscopy (T) and IB (U). (V–X) qPCR analysis of Ifnβ (V), Isg15 (W), and Cxcl10 (X) mRNA in Mavs^+/+ and Mavs^−/− MEF cells stably expressing empty vector or Sirt3, followed by transfection without (UT) or with high molecular weight poly I:C (poly I:C HMW) or low molecular weight poly I:C (poly I:C LMW) for 8 h. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data from at least three independent experiments (mean ± SD) (A–F, K–S, and V–X].

The viral RNA analog poly (I:C) LMW (low molecular weight) and HMW (high molecular weight) are the ligands for RIG-I and MDA5, respectively (Fig. 1J) (29). To determine which molecule is targeted by SIRT3, we used poly (I:C) LMW and HMW to activate RLR signaling respectively and then examined the effect of SIRT3 in this pathway. Overexpression of SIRT3 caused an increase of IFNβ mRNA upon both poly (I:C) LMW and HMW stimulation in H1299 cells (Fig. 1K), but disruption of SIRT3 in H1299 cells caused a reduction of IFNβ mRNA compared to that in WT H1299 cells (SIRT3^+/+) upon both poly (I:C) LMW and HMW stimulation (Fig. 1L). These data suggest that SIRT3 may target the merged molecule of the RIG-I pathway and MDA5 pathways, MAVS (Fig. 1J).

Overexpression of SIRT3 enhanced the luciferase activity of IFNβ promoter reporter (IFNβ-luc) and ISRE reporter (ISRE-luc) induced by ectopic expression of MAVS (Fig. 1 M and N). Consistent with these findings, overexpression of SIRT3 also promoted IFNβ mRNA expression induced by ectopic expression of MAVS (Fig. 1O). Furthermore, compared with those in Mavs-intact (Mavs^+/+) mouse embryonic fibroblast cells (MEF), overexpression of Sirt3 in Mavs-deficient MEF cells had no apparent effect on the induction of Ifnβ, Isg15, Cxcl10, and Ifit1 mRNA upon VSV infection (Fig. 1 P–S). Consistent with these observations, Sirt3 overexpression attenuated VSV-GFP virus propagation in Mavs^+/+ MEF cells, but Sirt3 overexpression had no apparent effect on VSV-GFP virus propagation in Mavs^−/− MEF cells, as shown by fluorescence microscopy imaging and immunoblotting (Fig. 1 T and U).

To exclude the possibility that these phenomena were due to a direct effect of Sirt3 on VSV replication, we further used the viral RNA analog poly (I:C) LMW and HMW to repeat these assays. As shown in Fig. 1 V–X, in Mavs^+/+ MEF cells, Sirt3 overexpression caused an increase of Ifnβ, Isg15, and Cxcl10 mRNA induced by the transfection with poly (I:C) LMW or HMW, whereas in Mavs^−/− MEF cells, Sirt3 overexpression had no apparent effect on the induction of Ifnβ, Isg15, and Cxcl10 mRNA.

Taken together, these data suggest that SIRT3 positively regulates RLR signaling by targeting MAVS.

SIRT3 Interacts with MAVS to Enhance MAVS Function Depending on Its Enzymatic Activity.

Notably, neither SIRT3 overexpression nor SIRT3 deletion affected MAVS protein levels (SI Appendix, Fig. S2 A and B). Furthermore, SIRT3 expression was not affected by VSV infection (SI Appendix, Fig. S2C).

The coimmunoprecipitation (Co-IP) assay showed that ectopically expressed MAVS interacted with ectopically expressed SIRT3 (Fig. 2 A and B), and endogenous MAVS and SIRT3 could also be coimmunoprecipitated (Fig. 2C). As expected, overexpressed SIRT3 was localized to mitochondria, and VSV infection promoted the colocalization of SIRT3 with TOM20, an outer mitochondrial membrane protein (SI Appendix, Fig. S3A). SIRT3 colocalized with MAVS and VSV infection enhanced their colocalization (Fig. 2D). The interaction between SIRT3 and MAVS was further confirmed by PLA (Fig. 2E). Due to the lack of a suitable anti-SIRT3 antibody that can detect endogenous SIRT3 by immunofluorescence staining (SI Appendix, Fig. S3 B–D), in order to mimic endogenous SIRT3 and obtain information about endogenous SIRT3 localization, we overexpressed SIRT3 at a relatively low dose in SIRT3^−/− H1299 cells to achieve comparable endogenous levels of SIRT3 in H1299 cells (SI Appendix, Fig. S3E) and then examined the colocalization between SIRT3 and MAVS. Similar to the above observations (SI Appendix, Fig. S3A and Fig. 2D), upon VSV infection, SIRT3 colocalized with TOM20 and MAVS (SI Appendix, Fig. S3 F and G). Domain mapping revealed that the TM domain of MAVS was required for SIRT3 binding (SI Appendix, Fig. S4 A–C).

Fig. 2.

SIRT3 interacts with MAVS to enhance MAVS activation depending on its enzymatic activity. (A and B) Co-IP of Flag-SIRT3 with HA-MAVS and vice versa. HEK293T cells were cotransfected with the indicated plasmids for 24 h. Anti-HA (A) or anti-Flag (B) antibody-conjugated agarose beads were used for IP, and the interaction was detected by immunoblotting with the indicated antibodies. (C) Endogenous interaction between MAVS and SIRT3. Anti-MAVS antibody was used for IP, and normal mouse IgG was used as a control. (D) Colocalization of SIRT3 and MAVS. H1299 cells were transfected with Flag-SIRT3 for 24 h, followed by UI or infected with VSV for 6 h. Confocal microscopy image of Flag-SIRT3 was detected by immunofluorescence staining with anti-Flag antibody, and endogenous MAVS was detected by immunofluorescence staining with anti-MAVS antibody, (Scale bar, 10 µm.) (E) H1299 cells were transfected with Flag-SIRT3 or Flag empty vector as control for 24 h, followed by UI (−) or infected with VSV (+) infection for 6 h. In situ PLAs of the SIRT3-MAVS interaction in H1299 cells with indicated combinations using anti-Flag and anti-MAVS antibodies. Quantification analysis of the SIRT3-MAVS interaction is shown in the Right figure. (Scale bar, 25 µm.) (F and G) qPCR analysis of IFNβ (F) and CXCL10 (G) mRNA in A549 cells stably expressing empty vector, SIRT3, or the enzymatically inactive mutant SIRT3-H248Y (lentivirus infection), followed by infection without (UI) or with VSV for 8 h. (H–J) A549 cells stably expressing empty vector, SIRT3, or the enzymatically inactive mutant SIRT3-H248Y were infected with VSV-GFP virus (MOI = 0.1) for 16 h, and viral infectivity was detected by flow cytometry analysis (n = 3) (H), fluorescence microscopy (I), or IB (J). (K and L) qPCR analysis of IFNβ (K) and CXCL10 (L) mRNA in SIRT3-deficient A549 cells (SIRT3^−/−) transfected with empty vector, Myc-tagged SIRT3 or the enzymatically inactive mutant SIRT3-H248Y for 24 h, followed by UI or infected with VSV for 8 h. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data from at least three independent experiments (mean ± SD) (E–H, K, and L).

In A549 cells, overexpression of WT SIRT3 enhanced VSV-induced expression of IFNβ and CXCL10, but ectopic expression of SIRT3-H248Y, an enzymatically inactive SIRT3 mutant (30), did not enhance VSV-induced expression of IFNβ and CXCL10 (Fig. 2 F and G). Consistent with these findings, overexpression of WT SIRT3 significantly suppressed VSV-GFP virus propagation, but ectopic expression of SIRT3-H248Y did not (Fig. 2 H–J). Similarly, reconstitution of WT SIRT3 enhanced VSV- or SeV-induced expression of IFNβ and CXCL10, but reconstitution of SIRT3-H248Y did not enhance VSV- or SeV-induced expression of IFNβ and CXCL10 (Fig. 2 K and L and SI Appendix, Fig. S4 D and E).

To further validate that the effect of SIRT3 on RLR signaling is dependent on its enzymatic activity, we took advantage of its specific inhibitor, 3-TYP (31). First, we confirmed that 3-TYP did not affect cell apoptosis and viability when used in the range of 10 to 200 μM (SI Appendix, Fig. S5 A and B). Treatment with 3-TYP suppressed the expression of IFNβ and CXCL10 in VSV-infected THP-1 cells (SI Appendix, Fig. S5 C and D). Similarly, 3-TYP treatment also suppressed the expression of IFNβ, CXCL10, and ISG15 in VSV-infected A549 cells (SI Appendix, Fig. S5 E–G). To exclude a direct effect of 3-TYP on viral replication, we repeated these assays using the viral RNA analog poly (I:C) LMW. Treatment with 3-TYP suppressed the expression of IFNβ and CXCL10 in poly (I:C) LMW-transfected H1299 cells (SI Appendix, Fig. S5 H and I). In SIRT3^−/− H1299 cells, treatment with 3-TYP had no apparent effect on the expression of IFNβ and CXCL10 in poly (I:C) LMW-transfected H1299 cells (SI Appendix, Fig. S5 H and I).

Taken together, these data suggest that SIRT3 binds to MAVS and inhibits MAVS activation in a manner dependent on its enzymatic activity.

SIRT3 Deacetylates MAVS at Lysine 7.

In HEK293T cells, addition of the histone deacetylase inhibitors trichostatin A (TSA) and nicotinamide (NAM) caused a time-dependent increase in acetylation of overexpressed Flag-MAVS, as revealed by anti-pan-acetyl-K antibody after IP with anti-Flag antibody (SI Appendix, Fig. S6A). We further confirmed this result after using anti-pan-acetyl-K antibody for IP (Fig. 3A). Next, we used MEF cells to test whether endogenous Mavs is also acetylated. Addition of TSA and NAM caused acetylation of endogenous Mavs in Mavs^+/+ MEF cells, but not in Mavs^−/− MEF cells (Fig. 3 B and C). Notably, acetylation of endogenous Mavs was higher in Sirt3^−/− MEF cells than in Sirt3^+/+ MEF cells, suggesting that Sirt3 may target Mavs to catalyze MAVS deacetylation (Fig. 3D).

Fig. 3.

SIRT3 deacetylates MAVS at lysine 7. (A) HEK293T cells were transfected with HA empty vector (HA empty) or HA-MAVS for 24 h. Anti-pan-acetyl-K antibody was used for IP, followed by immunoblotting with anti-MAVS antibody. (B) MEF cell lysates were immunoprecipitated with anti-MAVS antibody or mouse IgG control, followed by immunoblotting with anti-pan-acetyl-K antibody. (C) Mavs^−/− or Mavs^+/+ MEF cells were treated with (+) or without (−) TSA and NAM for 4 h, and the cell lysates were immunoprecipitated with anti-MAVS antibody, followed by immunoblotting with anti-pan-Acetyl-K antibody. (D) Cell lysates from Sirt3^−/− or Sirt3^+/+ MEF cells were immunoprecipitated with anti-MAVS antibody, followed by immunoblotting with anti-pan-Acetyl-K antibody. (E and F) qPCR analysis of IFNβ (E) and CXCL10 (F) mRNA in HEK293T cells transfected with HA-MAVS WT or mutants together with (+) or without (−) Flag-SIRT3 for 24 h. EV, empty vector. (G) Sequence alignment of partial MAVS (1 to 30 amino acids) from the human, macaque, cow, pig, dog, mouse, and rat. The red box indicates the conserved lysine 7 (K7). (H) Cell lysates from SIRT3^−/− or SIRT3^+/+ H1299 cells were immunoprecipitated with anti-MAVS antibody or mouse IgG control, followed by immunoblotting with anti-Ac-K7-MAVS antibody. (I) Cell lysates from SIRT3^−/− or SIRT3^+/+ A549 cells were immunoprecipitated with anti-MAVS antibody or mouse IgG control, followed by IB with anti-Ac-K7-MAVS antibody. (J) Cell lysates from A549 cells stably expressing empty vector, SIRT3, or the enzymatically inactive mutant SIRT3-H248Y were immunoprecipitated with anti-MAVS antibody or mouse IgG control, followed by immunoblotting with anti-Ac-K7-MAVS antibody. (K and L) Disruption of Sirt3 in mice increased MAVS acetylation in mouse spleen (K) and liver (L). Proteins extracted from spleens (K) and livers (L) of Sirt3^−/− or Sirt3^+/+ littermates (n = 5 per group) were analyzed with the indicated antibodies. MAVS acetylation was determined using anti-Ac-K7-MAVS antibody. Relative acetylation levels were quantified (Right). (M and N) qPCR analysis of IFNβ (M) and CXCL10 (N) mRNA in MAVS^−/− HEK293T cells transfected with HA-MAVS, or its acetylation-deficient mutant (HA-MAVS-K7R), or its acetylation mimic-mutant (HA-MAVS-K7Q or HA-MAVS-K7L) for 24 h. (O) IB of TBK1 and IRF3 phosphorylation in MAVS^−/− HEK293T cells transfected with HA-MAVS, or its acetylation-deficient mutant (HA-MAVS-K7R), or its acetylation mimic-mutant (HA-MAVS-K7Q or HA-MAVS-K7L), or its succinylation mimic-mutant (HA-MAVS-K7E) for 24 h. (P) SDD-AGE analysis of MAVS aggregates in MAVS^−/− HEK293T cells transfected with HA-MAVS or its acetylation-mimic mutant (HA-MAVS-K7Q or HA-MAVS-K7L) for 24 h, followed by infection with SeV for 12 h. SDS–PAGE immunoblotting was used as a loading control. ns, not significant (P > 0.05), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data from at least three independent experiments (mean ± SD) (E, F, and K–N).

As shown in Fig. 3 E and F, overexpression of SIRT3 had no enhancement effect on MAVS-K7R-induced expression of both IFNβ and CXCL10 in HEK293T cells, but still caused an increase in other mutant-induced expression of both IFNβ and CXCL10 in HEK293T cells. This lysine 7 is conserved not only between the mouse and human, but also between the human, macaque, cow, pig, dog, mouse, and rat (Fig. 3G). Thus, the conserved lysine 7 of MAVS may be specifically targeted by SIRT3 to mediate the enhancing effect of SIRT3 on RLR signaling (Fig. 3 E–G).

Subsequently, we developed a specific antibody against lysine 7 of MAVS (anti-Ac-K7-MAVS antibody). The specificity of this antibody was validated by dot blot assay (SI Appendix, Fig. S6B). When WT (HA-MAVS) or mutant MAVS (HA-MAVS-K7R) was ectopically expressed in HEK293T cells, only WT MAVS could be detected by the anti-Ac-K7-MAVS antibody in the precipitate obtained with anti-HA conjugated agarose beads (SI Appendix, Fig. S6C), confirming the specificity of the antibody. After Co-IP with anti-MAVS antibody, we found that MAVS acetylation at lysine 7 was much higher in SIRT3-deficient H1299 cells (SIRT3^−/−) than in WT H1299 cells (SIRT3^+/+), as revealed by anti-Ac-K7-MAVS antibody (Fig. 3H). Similar results were observed in SIRT3^+/+ and SIRT3^−/− A549 cells (Fig. 3I). Furthermore, in A549 cells, overexpression of WT SIRT3 (pHAGE-SIRT3-WT) reduced MAVS acetylation at lysine 7, whereas overexpression of the enzymatically inactive mutant of SIRT3 (pHAGE-SIRT3-H248Y) did not (Fig. 3J). As expected, the addition of 3-TYP restored the acetylation of MAVS (SI Appendix, Fig. S6 D and E).

To further determine the function of SIRT3 in vivo, we generated Sirt3-knockout mice using CRISPR/Cas9 technology (SI Appendix, Fig. S7A). Sirt3 deficiency in mice not only did not affect general development, but also did not affect immune cell development as previously reported (SI Appendix, Fig. S7 B and C) (32). However, MAVS acetylation at lysine 7 in the spleen and liver was significantly increased in Sirt3^−/− mice compared to Sirt3^+/+ mice as revealed by anti-Ac-K7-MAVS antibody, confirming the effect of Sirt3 on Mavs deacetylation at lysine 7 in vivo (Fig. 3 K and L). To validate the importance of lysine 7 acetylation in affecting MAVS activation, we generated two acetylation mimic mutants of MAVS, MAVS-K7L, and MAVS-K7Q (33). In MAVS^−/− HEK293T cells, the enhancement of MAVS-K7L on the expression of IFNβ and CXCL10 was significantly lower than that of WT MAVS (Fig. 3 M and N). Consistent with these findings, the phosphorylation of IRF3 (p-IRF3) and TBK1 (p-TBK1) induced by ectopic expression of MAVS-K7L and MAVS-K7Q in MAVS^−/− HEK293T cells was much weaker than that induced by ectopic expression of WT MAVS (Fig. 3O). The aggregates formed by MAVS-K7L and MAVS-K7Q were also much weaker than those formed by WT MAVS (Fig. 3P). Unexpectedly, the induction of IFNβ and CXCL10 expression and phosphorylation of IRF3 (p-IRF3) and TBK1 (p-TBK1), as well as aggregate formation, by ectopic expression of MAVS-K7R were weaker than those by ectopic expression of WT MAVS (Fig. 3 M–O and SI Appendix, Fig. S8A), which may be due to the simultaneous loss of multiple modifications in MAVS-K7R (15, 22, 34). In contrast to MAVS-K7L and MAVS-K7Q, phosphorylation of IRF3 (p-IRF3) and TBK1 (p-TBK1) and aggregate formation were much stronger upon ectopic expression of MAVS-K7E, a succinylation mimic mutant of MAVS, was much stronger (Fig. 3O and SI Appendix, Fig. S8B). In addition, SIRT3 overexpression or knockout had no obvious effect on SIRT5 protein levels (SI Appendix, Fig. S8 C and D).

These data suggest that SIRT3 targets MAVS at lysine 7 to deacetylate MAVS, leading to the enhancement of MAVS activation in RLR signaling.

Disruption of Sirt3 in Mice Attenuates the Cellular Antiviral Immune Response in MEF, BMDC, and BoneMarrow-Derived Macrophages (BMDM) Cells.

To determine the effects of Sirt3 on innate immunity in vivo, we used Sirt3-deficient mice. First, we established MEF cell lines from WT mice (Sirt3^+/+) and Sirt3-deficient mice (Sirt3^−/−) and challenged them with SeV or VSV. Compared to Sirt3^+/+ MEF cells, the expression of Ifnβ, Cxcl10, Cxcl11, Isg15, Ccl5, Rig-I, or Ifit1 mRNA was significantly reduced in Sirt3^−/− MEF cells upon SeV or VSV challenge (SI Appendix, Fig. S9 A–J). Consistent with this, Sirt3 deficiency attenuated the formation of Mavs aggregates in response to SeV infection (SI Appendix, Fig. S9K). Furthermore, phosphorylation of Irf3 and Tbk1 was impaired in Sirt3^−/− MEF cells compared to Sirt3^+/+ MEF cells upon VSV infection (SI Appendix, Fig. S9 L and M). In contrast, VSV-GFP virus propagation was enhanced in Sirt3^−/− MEF cells as shown by fluorescence microscopy imaging and immunoblotting (SI Appendix, Fig. S9N). Similarly, the expression of Ifnβ, Cxcl10, Isg15, or Ifit1 mRNA was significantly decreased in Sirt3^−/− MEF cells when transfected with poly (I:C) HMW or LMW compared with that in Sirt3^+/+ MEF cells (SI Appendix, Fig. S9 O–R). As expected, the Ifnβ levels were lower in Sirt3^−/− MEF cells compared to Sirt3^+/+ MEF cells after transfection with poly (I:C) LMW (SI Appendix, Fig. S9S). Similar results were obtained by infection with encephalomyocarditis virus (EMCV) (SI Appendix, Fig. S9 T–X). In Sirt3^−/− MEF cells, treatment with 3-TYP had no apparent effect on the expression of Ifnβ and Isg15 upon EMCV infection (SI Appendix, Fig. S9 Y and Z).

To further validate the above observations, we established both bone marrow-derived dendritic cells (BMDCs) and BMDMs and performed similar experiments. SeV- and VSV-induced expression of Ifnβ, Isg15, Ifit1, Rig-I, and Ccl5 mRNA was attenuated in Sirt3^−/− BMDC compared to Sirt3^+/+ BMDC upon SeV or VSV infection (SI Appendix, Fig. S10 A–J). Consistent with these findings, phosphorylation of Irf3 and Tbk1 was impaired in Sirt3^−/− BMDCs compared to that in Sirt3^+/+ BMDCs upon SeV or VSV infection (SI Appendix, Fig. S10 K–N). Similar results were obtained in Sirt3^+/+ and Sirt3^−/− BMDM cells upon SeV or VSV infection (SI Appendix, Fig. S10 O–Y).

Taken together, these data suggest that disruption of Sirt3 attenuates the cellular antiviral innate immune response.

Sirt3-Deficient Mice Are More Susceptible to Viral Infections.

After intraperitoneal injection (i.p.) of VSV, acetylation of MAVS in the spleen and liver of Sirt3^−/− mice was significantly higher than that of Sirt3^+/+ mice as detected by anti-Ac-K7-MAVS antibody (Fig. 4 A and B). As expected, the expression of Ifnβ, Ifnα1, Ifnα4, Cxcl10, Cxcl11, Ifit1, Isg15, or Rig-I mRNA was attenuated in the spleen, lung, and liver of Sirt3^−/− mice compared to those of Sirt3^+/+ mice after i.p. injection of VSV (Fig. 4 C–T).

Fig. 4.

Sirt3-deficient mice are more susceptible to viral infection. (A) Disruption of Sirt3 in mice increased MAVS acetylation in mouse spleen upon VSV infection. Proteins extracted from spleens of Sirt3^−/− and Sirt3^+/+ littermates injected intraperitoneally with VSV [3 × 10⁷ plaque-forming units (PFU) per mouse] for 24 h (n = 5 per group) were analyzed with the indicated antibodies. MAVS acetylation was detected by anti-Ac-K7-MAVS antibody and the relative acetylation level was quantified. (B) Disruption of Sirt3 in mice increased MAVS acetylation in mouse liver after VSV infection. Proteins extracted from livers of Sirt3^−/− and Sirt3^+/+ littermates injected intraperitoneally with VSV [3 × 10⁷ PFU per mouse] for 24 h (n = 5 per group) were analyzed with the indicated antibodies. MAVS acetylation was detected by anti-Ac-K7-MAVS antibody and the relative acetylation level was quantified. (C–H) qPCR analysis of Ifnβ (C), Ifnα1 (D), Ifnα4 (E), Cxcl10 (F), Cxcl11 (G), and Ifit1 (H) mRNA in the spleens of Sirt3^−/− and Sirt3^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (I–N) qPCR analysis of Ifnβ (I), Ifnα1 (J), Isg15 (K), Cxcl10 (L), Rig-I (M), and Ifit1 (N) mRNA in the lungs of Sirt3^−/− and Sirt3^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (O–T) qPCR analysis of Ifnβ (O), Ifnα1 (P), Isg15 (Q), Cxcl10 (R), Rig-I (S), and Ifit1 (T) mRNA in the livers of Sirt3^−/− and Sirt3^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (U) Survival (Kaplan–Meier curve) of Sirt3^−/− and Sirt3^+/+ mice (n = 8) injected intraperitoneally with a high dose of VSV (3 × 10⁷ PFU per mouse) and monitored for 10 d. *P < 0.05, using the log-rank test (Mantel-Cox). (V) ELISA of Ifnβ in serum from the Sirt3^−/− (n = 5) and Sirt3^+/+ mice (n = 5) injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (W) Hematoxylin-and-eosin-stained (H&E) images of lung sections from mice in (V). (Scale bar, 50 μm.) (X) Schematic of the working model for two RNA viruses, VSV and EMCV. (Y and Z) qPCR analysis of Ifnβ (Y) and Cxcl10 (Z) mRNA in the spleens of Sirt3^−/− and Sirt3^+/+ mice injected intraperitoneally with EMCV (1 × 10⁶ PFU per mouse) or PBS control for 24 h. ns, not significant (P > 0.05), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data from at least three independent experiments (mean ± SD) (A–T, V, Y, and Z).

Consistent with these findings, Sirt3^−/− mice were more susceptible to VSV infection than Sirt3^+/+ mice (Fig. 4U). Serum Ifnβ levels were lower in Sirt3^−/− mice than in Sirt3^+/+ mice (Fig. 4V). Hematoxylin and eosin (H&E) staining revealed more immune cell infiltration and injury in the lungs of Sirt3^−/− mice compared to Sirt3^+/+ mice (Fig. 4W). In contrast, VSV propagation was enhanced in Sirt3^−/− BMDC compared to Sirt3^+/+ BMDC (SI Appendix, Fig. S11 A and B). Similar results were observed in the spleen, liver, and lung of Sirt3^−/− mice compared to Sirt3^+/+ mice (SI Appendix, Fig. S11 C–E). In addition, the body weight of Sirt3^−/− mice was lower than that of Sirt3^+/+ mice after VSV infection for 6 d (SI Appendix, Fig. S11F).

Furthermore, after i.p. injection of EMCV, an ssRNA virus specifically sensed by MDA5 (Fig. 4X) (35), the expression of Ifnβ, Cxcl10, and Isg15 mRNA was reduced in the spleen, lung, and liver of Sirt3^−/− mice, compared to those of Sirt3^+/+ mice (Fig. 4 Y and Z and SI Appendix, Fig. S11 G–L).

These data suggest that Sirt3 is an important positive regulator of innate immunity against RNA viruses in vivo.

sirt3-Deficient Zebrafish Are More Susceptible to Viral Infection.

By comparing the amino acid sequences between different organisms, we found that SIRT3 is evolutionarily conserved, particularly in the C-terminus of SIRT3 (SI Appendix, Fig. S12A). To validate the general role of SIRT3 in antiviral innate immunity, we used the zebrafish model to perform additional assays (36). Upon infection of Epithelioma papulosum cyprini (EPC) cells with spring viremia of carp virus (SVCV, a single-strand RNA virus of fish) (37), overexpression of zebrafish sirt3 enhanced the expression of ifn, isg15, and viperin mRNA, similar to that was observed in mammalian models (SI Appendix, Fig. S12 B–D). In addition, zebrafish mavs interacted with zebrafish sirt3 when overexpressed (SI Appendix, Fig. S12E).

Using CRISPR/Cas9, we generated a mutant line with ten nucleotides deleted in exon 4 of zebrafish sirt3 (SI Appendix, Fig. S12F). After adding SVCV to the water containing zebrafish larvae (3 days post fertilization [dpf]) for 27 h, we found that sirt3-deficient zebrafish larvae (sirt3^−/−) were more susceptible to SVCV-induced death compared to WT zebrafish larvae (sirt3^+/+) (SI Appendix, Fig. S12 G and H). Similar results were obtained when WT zebrafish larvae (sirt3^+/+) and sirt3-deficient zebrafish larvae (sirt3^−/−) were infected with grass carp reovirus (GCRV), a double-stranded RNA virus (38) (SI Appendix, Fig. S12 I and J).

Next, we sought to determine whether zebrafish sirt3 function is dependent on its enzymatic activity using the SIRT3 inhibitor, 3-TYP. In zebrafish ZFL cells, upon SVCV infection or poly (I:C) transfection, the addition of 3-TYP significantly reduced the expression of ifnϕ1, lta, and mxc mRNA (SI Appendix, Fig. S13 A–I). Furthermore, in zebrafish larvae (3 dpf), the addition of 3-TYP also significantly reduced the expression of ifnϕ1, lta, and mxc mRNA upon SVCV infection (SI Appendix, Fig. S13 J–L). As expected, the addition of 3-TYP to the water containing zebrafish larvae (3 dpf) promoted SVCV-induced death (SI Appendix, Fig. S13 M and N).

Taken together, these data suggest that zebrafish sirt3 also positively regulates antiviral innate immunity, in a manner dependent on its enzymatic activity.

Disruption of Sirt5 in Sirt3-Deficient Mice Counteracts the Increased Viral Susceptibility Exhibited by Sirt3-Deficient Mice.

Interestingly, the same lysine of MAVS (lysine 7) can not only be modified simultaneously by SIRT3 and SIRT5, but also has opposing effects on MAVS activation (22). Notably, Sirt5^−/− mice are more resistant to viral infection (22), whereas Sirt3^−/− mice are more susceptible to viral infection. These phenomena led us to hypothesize that deacetylation of MAVS at lysine 7 by SIRT3 may act as an accelerator, but desuccinylation of MAVS at lysine 7 by SIRT5 may act as a brake in RLR signaling. To test this hypothesis, we generated Sirt3 and Sirt5 double knockout mice by crossing Sirt3^+/−Sirt5^+/− (♀) mice × Sirt3^+/−Sirt5^+/− (♂) mice (Fig. 5A). As previously reported, Sirt3^−/−Sirt5^−/− mice reproduced and developed without abnormalities, as well as their immune cell development was normal (SI Appendix, Fig. S14 A and B) (39). In BMDCs, SeV- and VSV-induced Ifnβ expression was quite similar between Sirt3^−/−Sirt5^−/− mice and their WT littermates (Sirt3^+/+Sirt5^+/+) (Fig. 5 B and C). Consistent with this, the level of Ifnβ in the supernatants was also similar between Sirt3^−/−Sirt5^−/− BMDC and their WT BMDCs (Sirt3^+/+Sirt5^+/+) (Fig. 5D).

Fig. 5.

Disruption of Sirt5 in Sirt3-deficient mice counteracts the viral susceptibility exhibited in Sirt3-deficient mice. (A) Schematic of generation of Sirt3 and Sirt5 double knockout mice (Sirt3^−/−Sirt5^−/−). (B) qPCR analysis of Ifnβ mRNA in Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ BMDCs infected without (0) or with SeV for 4 h or 8 h. (C) qPCR analysis of Ifnβ mRNA in Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ BMDCs infected without (0) or with VSV for 4 h or 8 h. (D) ELISA of Ifnβ in supernatants of Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ BMDCs infected with VSV for 12 h. (E) IB of Tbk1 and Irf3 phosphorylation in Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ BMDCs infected without (0) or with VSV for 4 h or 8 h. (F) Quantitation of phosphorylated Tbk1 (p-Tbk1) and Irf3 (p-Irf3) proteins shown in (E). (G–I) qPCR analysis of Ifnβ (G), Isg15 (H), and Cxcl10 (I) mRNA in the spleens of Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (J–L) qPCR analysis of Ifnβ (J), Isg15 (K), and Cxcl10 (L) mRNA in the lungs of Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (M–O) qPCR analysis of Ifnβ (M), Isg15 (N), and Cxcl10 (O) mRNA in the livers of Sirt3^−/−Sirt5^−/− or Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with VSV (3 × 10⁷ PFU per mouse) or PBS control for 24 h. (P–R) qPCR analysis of Ifnβ (P), Isg15 (Q), and Cxcl10 (R) mRNA in the spleens of Sirt3^−/−Sirt5^−/− and Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with EMCV (1 × 10⁶ PFU per mouse) or PBS control for 24 h. (S–U) qPCR analysis of Ifnβ (S), Isg15 (T), and Cxcl10 (U) mRNA in the lungs of Sirt3^−/−Sirt5^−/− and Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with EMCV (1 × 10⁶ PFU per mouse) or PBS control for 24 h. (V and W) qPCR analysis of Ifnβ (V) and Isg15 (W) mRNA in the livers of Sirt3^−/−Sirt5^−/− and Sirt3^+/+Sirt5^+/+ mice injected intraperitoneally with EMCV (1 × 10⁶ PFU per mouse) or PBS control for 24 h. (X) Survival (Kaplan–Meier curve) of Sirt3^−/−Sirt5 (n = 8) and Sirt3^+/+Sirt5^+/+ mice (n = 8) injected intraperitoneally with a high dose of EMCV (1 × 10⁶ PFU per mouse) and monitored for 6 d. ns, not significant (P > 0.05). (Y) The dynamic changes of acetylation and succinylation of MAVS at lysine 7 in response to VSV infection in Sirt3^+/+Sirt5^+/+, Sirt5^−/−, Sirt3^−/−, and Sirt3^−/−Sirt5^−/− MEF cells. MEF cells were infected with VSV for the indicated time points, and then cell lysates were immunoprecipitated with anti-MAVS antibody, followed by immunoblotting with the indicated antibodies. (Z) Quantitation of acetylated MAVS (Ac-K7-Mavs) and succinylated MAVS (Su-K7-Mavs) proteins shown in (Y). ns, not significant (P > 0.05). Data are from at least three independent experiments (mean ± SD) (B–D and G–W).

Furthermore, we observed that the phosphorylation of Irf3 and Tbk1 induced by VSV infection in BMDCs was almost the same between Sirt3^−/−Sirt5^−/− mice and their WT littermates (Sirt3^+/+Sirt5^+/+) (Fig. 5 E and F). Consistent with these findings, upon VSV infection, the expression of Ifnβ, Isg15, and Cxcl10 mRNA in the spleen, lung, and liver of Sirt3^−/−Sirt5^−/− mice was the same as that of Sirt3^+/+Sirt5^+/+ mice (Fig. 5 G–O).

In addition, upon EMCV infection, the expression of Ifnβ, Isg15, and Cxcl10 mRNA in the spleen, lung, and liver of Sirt3^−/−Sirt5^−/− mice was also the same as that of Sirt3^+/+Sirt5^+/+ mice (Fig. 5 P–W). As expected, EMCV-induced death was the same between Sirt3^−/−Sirt5^−/− mice and Sirt3^+/+Sirt5^+/+ mice (Fig. 5X). Taken together, these data suggest that the double knockout of Sirt5 and Sirt3 counteracts the enhancement of antiviral innate immunity by loss of Sirt5 as well as the impairment of antiviral innate immunity by loss of Sirt3, resulting in complete rescue of single mutant phenotypes exhibited in either Sirt3^−/− mice or Sirt5^−/− mice in response to viral infection.

This phenomenon prompted us to look closely at the underlying mechanisms. Notably, succinylation of lysine changes the charge of lysine from “+1” to “−1,” whereas acetylation of lysine changes the charge of lysine from “+1” to “0” (SI Appendix, Fig. S14C). It appears that succinylation of lysine reverses the charge in lysine, but acetylation of lysine neutralizes the charge in lysine. Thus, succinylation of the protein may strengthen its binding to its oppositely charged partners, but acetylation of the protein may weaken its binding to its partners due to its charge neutralization. Based on the structure of MAVS complexed with RIG-I, the tandem CARDs of RIG-I [2CARD (RIG-I)] tightly interact with the CARD of MAVS, leading to CARD (MAVS) filament assembly and activation of interferon signaling (SI Appendix, Fig. S14 D and E) (40). Structural modeling of MAVS succinylation and acetylation at lysine 7 shows that acetylation at K7 results in a longer distance between MAVS and the components of the MAVS/RIG-I complex (9.4, 6.8, and 7.5 versus 3.9, 6.8, and 6.7) compared to succinylation at K7 (SI Appendix, Fig. S14 D and E). Therefore, acetylation of MAVS at lysine 7 may weaken its binding to RIG-I, resulting in impaired RLR signaling.

In WT MEF cells, MAVS acetylation and succinylation were maintained at relatively uniform levels upon VSV infection (Fig. 5 Y and Z). In Sirt5^−/− MEF cells, upon VSV infection, MAVS succinylation was enhanced, but MAVS acetylation was maintained at a very low level (Fig. 5 Y and Z). Conversely, in Sirt3^−/− MEF cells, upon VSV infection, MAVS acetylation was enhanced, but MAVS succinylation remained at a very low level (Fig. 5 Y and Z). However, upon VSV infection, it appeared that both MAVS acetylation and MAVS succinylation in Sirt3^−/−Sirt5^−/− MEF cells were similar to those in Sirt3^+/+Sirt5^+/+MEF cells (Fig. 5 Y and Z). These data suggest that MAVS acetylation and succinylation at lysine 7 might interfere with each other.

Based on our observations, we proposed a working model for the control of MAVS activation by SIRT3-mediated deacetylation and SIRT5-mediated desuccinylation (SI Appendix, Fig. S15).

Discussion

In this study, we show that two different modifications on the same lysine of MAVS have opposing effects on MAVS-mediated antiviral innate immunity, revealing a previously unrecognized mechanism for orchestrating innate immunity. However, how these two modifications dynamically change and cooperate to exert their functions in maintaining immune homeostasis during both resting states and pathogen infection remains a mystery. Of note, it appears that acetylation can interfere with succinylation on lysine 7 of MAVS and vice versa during viral infection, thus, the dual modification on lysine 7 of MAVS may cooperate to regulate MAVS function in the RLR signaling.

As reported, SIRT3 is a primarily mitochondrial matrix protein and lysine 7 of MAVS is in the cytoplasm (41–44). If SIRT3 can modify MAVS, then lysine 7 of MAVS should be accessible to SIRT3. Although we have tested several commercially available anti-SIRT3 antibodies, none of them can detect endogenous SIRT3 localization by immunofluorescence staining. Thus, we could not directly observe that endogenous SIRT3 can colocalize with endogenous MAVS. In this study, we show that endogenous SIRT3 coimmunoprecipitates with endogenous MAVS. In addition, we show that overexpressed SIRT3 at comparable endogenous levels colocalizes with TOM2O, an outer mitochondrial membrane protein upon viral infection. Furthermore, SIRT3 colocalizes with MAVS during viral infection. Indeed, SIRT3 has been reported to localize to other parts of the cell in addition to the mitochondrial matrix under various circumstances. Thus, SIRT3 may translocate to the outer mitochondrial membrane to interact with and modify MAVS in response to viral infection. Similar behaviors may also exist for SIRT5, which is predominantly localized to the mitochondria, but also exerts regulatory activity in the cytoplasm (45–47).

The N-terminal CARD (from 10 to 93 amino acids) is the key domain for MAVS aggregation and subsequent activation (44, 48, 49). However, lysine 7 is close to the CARD, but not exactly localized to it. Both deacetylation and desuccinylation of MAVS at lysine 7 can dramatically affect its aggregation. Therefore, lysine 7 of MAVS may be an important site for determining MAVS function, even though it does not localize to the CARD. Indeed, other modifications at lysine 7 of MAVS that affect MAVS function have been previously identified (15, 34, 50–52). Structural modeling indicates that, compared to succinylation of MAVS at lysine 7, acetylation of MAVS at lysine 7 can increase the distance between MAVS and the components of the MAVS/RIG-I complex, leading to attenuation of MAVS-RIG-I complex formation and subsequent RLR signaling activation, further highlighting that the lysine 7 of MAVS is a critical residue for controlling MAVS function.

Notably, other residues of MAVS besides lysine 7 have been reported to be modified by various PTMs (19). TRIM25 or MARCH5 catalyzes K48-linked ubiquitination at K7, leading to the destabilization of MAVS or reduction of MAVS aggregates (15, 34). Succinylation at K7 activates MAVS, while desuccinylation at K7 by SIRT5 suppresses MAVS-mediated RLR signaling. Acetylation at K7 inhibits MAVS, but deacetylation at K7 by SIRT3 augments MAVS-mediated RLR signaling. When K7 of MAVS is mutated to arginine, the ubiquitination, succinylation, acetylation, and some other unspecified modifications at K7 are simultaneously blocked. It remains unclear to what extent deacetylation/desuccinylation of MAVS at lysine 7 contributes to the regulation of MAVS function in the RLR signaling, as determined by the use of the K7R mutant. Further investigation of this question may open a window for understanding the physiological function of MAVS deacetylation/desuccinylation in type I interferon signaling.

While the in vitro data obtained in this study support a role for SIRT3 in the regulation of MAVS-dependent immunity, the in vivo data based on Sirt3 knockout mice do not prove that SIRT3 acts exclusively through MAVS. In the future, obtaining Sirt3/Mavs double knockout mice will allow us to define MAVS-dependent and independent effects of SIRT3 in antiviral immunity.

Materials and Methods

The following is a brief description of Materials and Methods, and the description of Materials and Methods in detail in SI Appendix.

Mice.

Six- to eight-week-old mice were used in the experiments. Same-sex (male) littermates were randomly assigned to the experimental groups. All analyses were blinded.

Zebrafish.

CRISPR/Cas9 was used to generate sirt3 knockout zebrafish. Sirt3 sgRNA was designed using an online CRISPR design tool (http://crispor.gi.ucsc.edu/crispor.py).

Cell Culture.

Primary MEFs were prepared from embryos at 12.5 to 14.5 embryonic days. Bone marrow cells were isolated from the mouse femur.

Luciferase Reporter Assays.

Luciferase activity was determined using the dual-luciferase reporter assay system (Promega) at 18 to 24 h posttransfection.

Quantitative Real-Time PCR (qPCR).

MonAmp^TM SYBR^® Green qPCR Mix (high Rox) was used for qPCR assays. Primers for qPCR are listed in SI Appendix, Table S2.

CRISPR-Cas9 Knockout Cell Lines.

sgRNA sequence was ligated into LentiCRISPRv2 plasmid and then cotransfected with viral packaging plasmids (psPAX2 and pMD2G) into HEK293T cells.

Immunoblotting and IP Assay.

Cell lysates were separated by SDS-PAGE, transferred onto a polyvinylidene difluoride membrane, blocked with 5% (w/v) nonfat milk, probed with the indicated primary antibodies and corresponding secondary antibodies.

Immunofluorescence Confocal Microscopy.

Cells were stained with the indicated primary antibodies followed by incubation with fluorescent dye-conjugated secondary antibodies.

Proximity Ligation Assay (PLA).

PLA was performed according to the manufacturer’s instructions for the Duolink In Situ Red Starter Kit Mouse/Rabbit.

Generation of Anti-Ac-K7-MAVS Antibody.

A MAVS K7 site-specific acetylation antibody (anti-Ac-K7-MAVS) was generated by using a human MAVS acetylated peptide [MPFAED(ac-K)TYKYIC] as an antigen.

Semidenaturing Detergent Agarose Gel Electrophoresis (SDD-AGE) Assay.

Crude mitochondrial and cytosolic extracts were prepared from HEK293T or MEF cells by differential centrifugation.

Flow Cytometry Assays for VSV-GFP Infected Cells.

Cells were infected with VSV-GFP virus (MOI = 0.1) for the indicated time, then harvested and washed with PBS. Cells were counted at 10,000 cells, and subsequent analyses were performed on Beckman CytoFLEXS.

Cell Apoptosis Analysis.

FITC Annexin V Apoptosis Detection Kit I (BD Pharmingen) was used to measure the effect of 3-TYP (10 to 200 μM) on cell apoptosis of THP-1 cells.

Lentivirus-Mediated Gene Transfer.

Cells were transfected with pHAGE-SIRT3, pHAGE-SIRT3-H248Y, pHAGE-Sirt3, or the empty vector together with the packaging vectors pSPAX2 and pMD2G.

Structure Modeling.

We used PyMOL (http://www.pymol.org) to obtain the structure model of K7 acetylation and succinylation.

Virus Infection in Mice.

Age- and sex-matched (male) Sirt3^+/+ and Sirt3^−/− mice were infected with VSV. For viral infection in Sirt^+/+Sirt5^+/+ and Sirt3^−/−Sirt5^−/− mice, age- and sex-matched (male) mice were injected intraperitoneally with VSV or EMCV.

Viral Infection in Zebrafish.

Zebrafish larvae (3 dpf) were placed in a disposable 60-mm cell culture dish filled with 3 mL egg water and 2 mL SVCV (~2.0 × 10⁸ TCID₅₀/mL) culture medium, or 4.7 mL egg water and 0.3 mL GCRV-II.

Statistical Analysis.

GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA) was used for all statistical analyses.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.